Skip to content

very low throughputs of Cassandra for YCSB workload for large amount of data #148

Closed
prashant23 opened this Issue Oct 29, 2013 · 2 comments

4 participants

@prashant23

Hi all
Am benchmarking cassandra cluster using YCSB tool. When am using 1 gb data it is giving me fair results (throughput around 20K op/sec) but when am using the 1TB data then it is coming 200 to 300 op/sec. it is having very high read latency can any body tell me that what can be done to improve the throughput.
Thanks

@lambdaupb

You are probably hitting the Disks.

A 1GB Dataset fits into the Disk Cache of the OS.

With 1TiB this is no longer the case and most of the accesses need to do a seek on the Disk.
Depending on which type of Disk you use (HDD/SSD) this can take ~10ms per seek.

SSDs are much better at this.

@busbey
Collaborator
busbey commented May 29, 2015

closing as stale. if you are still having this issue please reopen.

@busbey busbey closed this May 29, 2015
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Something went wrong with that request. Please try again.